R Notebook

Loading necessary libraries

library(tidyverse)

Registered S3 methods overwritten by 'dbplyr':
  method         from
  print.tbl_lazy     
  print.tbl_sql      
── Attaching packages ──────────────────────────────────────────────────────────────────────────────────────────── tidyverse 1.3.1 ──
✔ ggplot2 3.4.4     ✔ purrr   1.0.2
✔ tibble  3.2.1     ✔ dplyr   1.1.4
✔ tidyr   1.3.0     ✔ stringr 1.5.1
✔ readr   2.1.4     ✔ forcats 1.0.0
── Conflicts ─────────────────────────────────────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()

library(visdat)
library(missForest)
library(corrplot)

corrplot 0.92 loaded

library(mice)

Attaching package: ‘mice’

The following object is masked from ‘package:stats’:

    filter

The following objects are masked from ‘package:base’:

    cbind, rbind

library(car)

Loading required package: carData

Attaching package: ‘car’

The following object is masked from ‘package:dplyr’:

    recode

The following object is masked from ‘package:purrr’:

    some

library("MVA")

Loading required package: HSAUR2
Loading required package: tools

Attaching package: ‘HSAUR2’

The following object is masked from ‘package:mice’:

    toenail

The following object is masked from ‘package:tidyr’:

    household

library(tidyverse)
library(psych)

Attaching package: ‘psych’

The following object is masked from ‘package:car’:

    logit

The following objects are masked from ‘package:ggplot2’:

    %+%, alpha

library(naniar)

options(scipen=999)

Combining two sets of data

data <- read.csv("data.csv")

data

Add region categorical variable

library(countrycode)
countries <- c(data$country)
data$region <- countrycode(sourcevar = countries,
                            origin = "country.name",
                            destination = "region")

Warning: Some values were not matched unambiguously: European Union

library(choroplethr)

Loading required package: acs
Loading required package: XML

Attaching package: ‘XML’

The following object is masked from ‘package:tools’:

    toHTML


Attaching package: ‘acs’

The following object is masked from ‘package:dplyr’:

    combine

The following object is masked from ‘package:base’:

    apply

Registered S3 method overwritten by 'htmlwidgets':
  method           from         
  print.htmlwidget tools:rstudio
Registered S3 method overwritten by 'data.table':
  method           from
  print.data.table     
Stuck? Ask a question here: https://stackoverflow.com/questions/tagged/choroplethr.

library(WDI) 

choroplethr_wdi(code="SH.STA.MMRT", year=2018, title="2018 Maternal Mortality Ratio", num_colors=1)

Warning: Please use ISO-2, ISO-3, or World Bank regional codes. Some of the country codes that you requested are invalid: NA, EH, AQWarning: The following regions were missing and are being set to NA: namibia, western sahara, taiwan, antarctica, kosovo

choroplethr_wdi(code="SP.DYN.TFRT.IN", year=2018, title="2018 Total Fertility Rate", num_colors=1)

Warning: Please use ISO-2, ISO-3, or World Bank regional codes. Some of the country codes that you requested are invalid: NA, EH, AQWarning: The following regions were missing and are being set to NA: namibia, western sahara, taiwan, antarctica, kosovo

choroplethr_wdi(code="SH.XPD.PVTD.PC.CD", year=2018, title="2018 Private Health Expenditure", num_colors=9)

Warning: Please use ISO-2, ISO-3, or World Bank regional codes. Some of the country codes that you requested are invalid: NA, EH, AQWarning: The following regions were missing and are being set to NA: namibia, north korea, western sahara, somaliland, somalia, syria, taiwan, yemen, antarctica, kosovo, libya

Remove rows that have more than 50% missing data.

final_df <- data[rowSums(is.na(data)) < ncol(data)/2, ]
final_df

miss_var_summary(final_df)

vis_miss(final_df)

Imputing missing data

categorical <- subset(data_after_mice, select = c(region, country, country_code))
non_categorical <- subset(data_after_mice, select = -c(region, country, country_code))
res <- cor(non_categorical)
corrplot(res, type = "upper", order = "hclust", 
         tl.col = "black", tl.srt = 45,col = COL2('PuOr', 10))

Remove variables that present multicollinearity

no_controls = subset(data_after_mice, select = -c(region, country, country_code))

vif_fun <- function(df, keep_in) {
             # df: the dataset of interest
             # keep_in: the variables that should be kept in  
             highest <- c()
             while(TRUE) {
                # the rnorm() below is arbitrary as the VIF should not 
                # depend on it
                vifs <- vif(lm(rnorm(nrow(df)) ~. , data = df))
                adj_vifs <- vifs[-which(names(vifs) %in% keep_in)]
                if (max(adj_vifs) < 10) {
                     break
                }
               highest <- c(highest,names((which(adj_vifs == max(adj_vifs)))))
               cat("\n")
               cat("removed:", highest)
               cat("\n")
               df <- df[,-which(names(df) %in% highest)]

              }
            cat("\n")
            cat("final variables: \n")
            return(names(vifs))
              }

vif_fun(no_controls,keep_in = c("gdp","fertility_rate","skilled_health_staff","mmr"))

removed: total_health_exp

removed: total_health_exp oop_health_exp

final variables: 
 [1] "govt_health_exp"    "comp_educ"          "mmr"                "gdp"                "gdp_per_capita"     "private_health_exp"
 [7] "fertility_rate"     "sab"                "num_physicians"     "rule_of_law"        "female_pop"        

data_after_vif <- subset(data_after_mice, select = -c(total_health_exp, oop_health_exp))
noncategorical <- subset(data_after_vif, select = -c(region, country, country_code))
boxplot(noncategorical, use.cols = TRUE, las=2)

for (i in 1:ncol(noncategorical)){
  plot(noncategorical[,i], noncategorical$mmr,
   xlab=colnames(noncategorical)[i], ylab="MMR", pch=19)
}

Remove export_val_ind outliers

data_transformed <- data_after_mice

data_transformed$gdp_transform <- log(data_after_mice$gdp)
data_transformed$fertility_rate_transform <- log(data_after_mice$fertility_rate)
data_transformed$mmr_transform <- log(data_after_mice$mmr)
data_transformed$female_pop_transform <- log(data_after_mice$female_pop)
data_transformed$govt_health_transform <- log(data_after_mice$govt_health_exp)
data_transformed$gpd_pc_transform <- log(data_after_mice$gdp_per_capita)
data_transformed$private_health_transform <- log(data_after_mice$private_health_exp)

data_after_mice$gdp_transform <- log(data_after_mice$gdp)
data_after_mice$fertility_rate_transform <- log(data_after_mice$fertility_rate)
data_after_mice$mmr_transform <- log(data_after_mice$mmr)
data_after_mice$female_pop_transform <- log(data_after_mice$female_pop)
data_after_mice$govt_health_transform <- log(data_after_mice$govt_health_exp)
data_after_mice$gpd_pc_transform <- log(data_after_mice$gdp_per_capita)
data_after_mice$private_health_transform <- log(data_after_mice$private_health_exp)

data_transformed <- subset(data_transformed, select = -c(female_pop, fertility_rate, mmr, gdp, govt_health_exp, gdp_per_capita, private_health_exp))

transformed_noncat <- subset(data_transformed, select = -c(region, country, country_code))

for (i in 1:ncol(transformed_noncat)){
  plot(transformed_noncat[,i], transformed_noncat$mmr,
   xlab=colnames(transformed_noncat)[i], ylab="MMR", pch=19)
}

boxplot(transformed_noncat, use.cols = TRUE, las=2)

baseline <- lm(mmr_transform ~ govt_health_transform + private_health_transform, data = data_after_mice)
summary(baseline)

Call:
lm(formula = mmr_transform ~ govt_health_transform + private_health_transform, 
    data = data_after_mice)

Residuals:
    Min      1Q  Median      3Q     Max 
-3.6261 -0.4506  0.0616  0.5454  2.0302 

Coefficients:
                         Estimate Std. Error t value             Pr(>|t|)    
(Intercept)               7.68130    0.20453  37.556 < 0.0000000000000002 ***
govt_health_transform    -0.51465    0.05968  -8.624  0.00000000000000208 ***
private_health_transform -0.26739    0.07659  -3.491             0.000593 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.9053 on 198 degrees of freedom
Multiple R-squared:  0.7135,    Adjusted R-squared:  0.7106 
F-statistic: 246.6 on 2 and 198 DF,  p-value: < 0.00000000000000022

baseline <- lm(mmr_transform ~ region + govt_health_transform + private_health_transform, data = data_after_mice)
summary(baseline)

Call:
lm(formula = mmr_transform ~ region + govt_health_transform + 
    private_health_transform, data = data_after_mice)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.73803 -0.35778  0.04387  0.48433  2.35595 

Coefficients:
                                 Estimate Std. Error t value             Pr(>|t|)    
(Intercept)                       6.52269    0.23691  27.532 < 0.0000000000000002 ***
regionEurope & Central Asia      -0.83576    0.17440  -4.792        0.00000330545 ***
regionLatin America & Caribbean   0.60639    0.18320   3.310             0.001115 ** 
regionMiddle East & North Africa -0.27143    0.20385  -1.332             0.184595    
regionNorth America               0.41232    0.45156   0.913             0.362335    
regionSouth Asia                  0.27076    0.29298   0.924             0.356562    
regionSub-Saharan Africa          1.00278    0.18587   5.395        0.00000020109 ***
govt_health_transform            -0.33308    0.05293  -6.293        0.00000000208 ***
private_health_transform         -0.24349    0.06486  -3.754             0.000231 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.7087 on 191 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8295,    Adjusted R-squared:  0.8223 
F-statistic: 116.1 on 8 and 191 DF,  p-value: < 0.00000000000000022

data_after_mice

baseline_contr_1 <- lm(mmr_transform ~ region + govt_health_transform + private_health_transform + gpd_pc_transform + gdp_transform + rule_of_law + comp_educ + female_pop_transform + fertility_rate_transform + sab, data = data_after_mice)
summary(baseline_contr_1)

Call:
lm(formula = mmr_transform ~ region + govt_health_transform + 
    private_health_transform + gpd_pc_transform + gdp_transform + 
    rule_of_law + comp_educ + female_pop_transform + fertility_rate_transform + 
    sab, data = data_after_mice)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.55146 -0.33833  0.02559  0.41833  2.39345 

Coefficients:
                                  Estimate Std. Error t value    Pr(>|t|)    
(Intercept)                       6.488229   1.108982   5.851 0.000000022 ***
regionEurope & Central Asia      -0.796719   0.169876  -4.690 0.000005311 ***
regionLatin America & Caribbean   0.608830   0.196094   3.105    0.002205 ** 
regionMiddle East & North Africa -0.410757   0.207745  -1.977    0.049511 *  
regionNorth America               0.197179   0.427661   0.461    0.645297    
regionSouth Asia                  0.259955   0.278046   0.935    0.351048    
regionSub-Saharan Africa          0.573976   0.199503   2.877    0.004489 ** 
govt_health_transform            -0.210971   0.072638  -2.904    0.004130 ** 
private_health_transform         -0.191597   0.075296  -2.545    0.011763 *  
gpd_pc_transform                  0.241784   0.194977   1.240    0.216531    
gdp_transform                    -0.244131   0.168079  -1.452    0.148072    
rule_of_law                      -0.017058   0.097380  -0.175    0.861137    
comp_educ                         0.020288   0.021599   0.939    0.348802    
female_pop_transform              0.265336   0.167947   1.580    0.115852    
fertility_rate_transform          0.730298   0.216996   3.365    0.000930 ***
sab                              -0.017295   0.004803  -3.601    0.000407 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.6637 on 184 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8559,    Adjusted R-squared:  0.8442 
F-statistic: 72.86 on 15 and 184 DF,  p-value: < 0.00000000000000022

baseline_contr_2 <- lm(mmr_transform ~ region + govt_health_transform + private_health_transform + gdp_transform + fertility_rate_transform + sab, data = data_after_mice)
summary(baseline_contr_2)

Call:
lm(formula = mmr_transform ~ region + govt_health_transform + 
    private_health_transform + gdp_transform + fertility_rate_transform + 
    sab, data = data_after_mice)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.52635 -0.35056  0.03425  0.43336  2.47912 

Coefficients:
                                  Estimate Std. Error t value          Pr(>|t|)    
(Intercept)                       6.077715   0.788813   7.705 0.000000000000731 ***
regionEurope & Central Asia      -0.769487   0.163740  -4.699 0.000005029707029 ***
regionLatin America & Caribbean   0.687525   0.173052   3.973          0.000101 ***
regionMiddle East & North Africa -0.416990   0.195870  -2.129          0.034564 *  
regionNorth America               0.234254   0.424062   0.552          0.581326    
regionSouth Asia                  0.269877   0.276803   0.975          0.330825    
regionSub-Saharan Africa          0.530399   0.196446   2.700          0.007567 ** 
govt_health_transform            -0.227863   0.053658  -4.247 0.000034093025032 ***
private_health_transform         -0.191903   0.069811  -2.749          0.006563 ** 
gdp_transform                     0.022464   0.024036   0.935          0.351194    
fertility_rate_transform          0.740756   0.210669   3.516          0.000549 ***
sab                              -0.015892   0.004669  -3.404          0.000812 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.663 on 188 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8531,    Adjusted R-squared:  0.8445 
F-statistic: 99.25 on 11 and 188 DF,  p-value: < 0.00000000000000022

data_after_mice[data_after_mice$region == "East Asia & Pacific",]

NA

data_after_mice[data_after_mice$region == "Europe & Central Asia",]

NA

data_after_mice$spend_ratio <- data_after_mice$govt_health_transform/data_after_mice$private_health_transform

ratio_lm <- lm(mmr_transform ~ region + spend_ratio + govt_health_transform + private_health_transform + gdp_transform + fertility_rate_transform + sab, data = data_after_mice)
summary(ratio_lm)

Call:
lm(formula = mmr_transform ~ region + spend_ratio + govt_health_transform + 
    private_health_transform + gdp_transform + fertility_rate_transform + 
    sab, data = data_after_mice)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.51519 -0.35391  0.03489  0.44213  2.42406 

Coefficients:
                                  Estimate Std. Error t value    Pr(>|t|)    
(Intercept)                       5.586261   0.927840   6.021 0.000000009 ***
regionEurope & Central Asia      -0.736182   0.167049  -4.407 0.000017637 ***
regionLatin America & Caribbean   0.716565   0.175439   4.084 0.000065470 ***
regionMiddle East & North Africa -0.378378   0.199590  -1.896    0.059532 .  
regionNorth America               0.240026   0.424087   0.566    0.572084    
regionSouth Asia                  0.308339   0.279423   1.103    0.271236    
regionSub-Saharan Africa          0.594936   0.206651   2.879    0.004455 ** 
spend_ratio                       0.325085   0.323178   1.006    0.315763    
govt_health_transform            -0.319462   0.105694  -3.023    0.002858 ** 
private_health_transform         -0.085852   0.126446  -0.679    0.498003    
gdp_transform                     0.025871   0.024273   1.066    0.287872    
fertility_rate_transform          0.728524   0.211013   3.453    0.000687 ***
sab                              -0.015959   0.004669  -3.418    0.000775 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.663 on 187 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8539,    Adjusted R-squared:  0.8445 
F-statistic: 91.07 on 12 and 187 DF,  p-value: < 0.00000000000000022

no_controls = subset(data_after_mice, select = -c(region, country, country_code))

vif_fun <- function(df, keep_in) {
             # df: the dataset of interest
             # keep_in: the variables that should be kept in  
             highest <- c()
             while(TRUE) {
                # the rnorm() below is arbitrary as the VIF should not 
                # depend on it
                vifs <- vif(lm(rnorm(nrow(df)) ~. , data = df))
                adj_vifs <- vifs[-which(names(vifs) %in% keep_in)]
                if (max(adj_vifs) < 10) {
                     break
                }
               highest <- c(highest,names((which(adj_vifs == max(adj_vifs)))))
               cat("\n")
               cat("removed:", highest)
               cat("\n")
               df <- df[,-which(names(df) %in% highest)]

              }
            cat("\n")
            cat("final variables: \n")
            return(names(vifs))
              }

vif_fun(no_controls,keep_in = c("gdp","fertility_rate","skilled_health_staff","mmr"))

removed: total_health_exp

removed: total_health_exp gdp_transform

removed: total_health_exp gdp_transform govt_health_transform

removed: total_health_exp gdp_transform govt_health_transform fertility_rate_transform

removed: total_health_exp gdp_transform govt_health_transform fertility_rate_transform gpd_pc_transform

removed: total_health_exp gdp_transform govt_health_transform fertility_rate_transform gpd_pc_transform oop_health_exp

final variables: 
 [1] "govt_health_exp"          "comp_educ"                "mmr"                      "gdp"                     
 [5] "gdp_per_capita"           "private_health_exp"       "fertility_rate"           "sab"                     
 [9] "num_physicians"           "rule_of_law"              "female_pop"               "mmr_transform"           
[13] "female_pop_transform"     "private_health_transform" "spend_ratio"             

vifLm <- lm(mmr_transform~region + govt_health_transform + private_health_transform + gdp_transform + fertility_rate_transform + sab, data = data_after_mice)
plot(vifLm)

dfNoOutliers <- data_after_mice[-c(49,74,19),]

no_outliers_lm <- lm(mmr_transform ~ region + govt_health_transform + private_health_transform + gdp_transform + fertility_rate_transform + sab, data = dfNoOutliers)

summary(no_outliers_lm)

Call:
lm(formula = mmr_transform ~ region + govt_health_transform + 
    private_health_transform + gdp_transform + fertility_rate_transform + 
    sab, data = dfNoOutliers)

Residuals:
     Min       1Q   Median       3Q      Max 
-2.30588 -0.36768  0.02031  0.42348  1.32428 

Coefficients:
                                  Estimate Std. Error t value            Pr(>|t|)    
(Intercept)                       6.055698   0.701913   8.627 0.00000000000000281 ***
regionEurope & Central Asia      -0.843949   0.148393  -5.687 0.00000004966615839 ***
regionLatin America & Caribbean   0.606159   0.156173   3.881            0.000144 ***
regionMiddle East & North Africa -0.507917   0.175939  -2.887            0.004354 ** 
regionNorth America               0.162204   0.377268   0.430            0.667736    
regionSouth Asia                  0.188490   0.245688   0.767            0.443945    
regionSub-Saharan Africa          0.425876   0.176517   2.413            0.016812 *  
govt_health_transform            -0.233037   0.047900  -4.865 0.00000243956902508 ***
private_health_transform         -0.184430   0.064100  -2.877            0.004483 ** 
gdp_transform                     0.022147   0.021496   1.030            0.304221    
fertility_rate_transform          0.791886   0.189802   4.172 0.00004635612096637 ***
sab                              -0.015200   0.004144  -3.668            0.000319 ***
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.5869 on 185 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8817,    Adjusted R-squared:  0.8747 
F-statistic: 125.3 on 11 and 185 DF,  p-value: < 0.00000000000000022

ratio_lm <- lm(mmr_transform ~ region + spend_ratio + gpd_pc_transform + gdp_transform + rule_of_law + comp_educ + female_pop_transform + fertility_rate_transform + sab, data = dfAfterVIF)

Error in is.data.frame(data) : object 'dfAfterVIF' not found

interaction_lm <- lm(mmr_transform ~ region + govt_health_transform + private_health_transform + gdp_transform + fertility_rate_transform + sab + region*govt_health_transform + region*private_health_transform, data = data_after_mice)
summary(interaction_lm)

Call:
lm(formula = mmr_transform ~ region + govt_health_transform + 
    private_health_transform + gdp_transform + fertility_rate_transform + 
    sab + region * govt_health_transform + region * private_health_transform, 
    data = data_after_mice)

Residuals:
    Min      1Q  Median      3Q     Max 
-2.5588 -0.2828  0.0000  0.3913  1.9616 

Coefficients:
                                                            Estimate Std. Error t value            Pr(>|t|)    
(Intercept)                                                 7.154288   0.837772   8.540 0.00000000000000617 ***
regionEurope & Central Asia                                -3.799619   0.659677  -5.760 0.00000003685332698 ***
regionLatin America & Caribbean                             0.126602   0.802424   0.158            0.874815    
regionMiddle East & North Africa                           -0.893621   0.703971  -1.269            0.205974    
regionNorth America                                        37.722845 102.900300   0.367            0.714361    
regionSouth Asia                                           -1.841573   1.274540  -1.445            0.150266    
regionSub-Saharan Africa                                   -0.992494   0.520110  -1.908            0.057988 .  
govt_health_transform                                      -0.311804   0.103018  -3.027            0.002844 ** 
private_health_transform                                   -0.416170   0.106797  -3.897            0.000138 ***
gdp_transform                                               0.030034   0.024604   1.221            0.223828    
fertility_rate_transform                                    0.489113   0.215709   2.267            0.024578 *  
sab                                                        -0.011995   0.004720  -2.541            0.011912 *  
regionEurope & Central Asia:govt_health_transform          -0.085542   0.147237  -0.581            0.561996    
regionLatin America & Caribbean:govt_health_transform       0.074779   0.201933   0.370            0.711593    
regionMiddle East & North Africa:govt_health_transform      0.113804   0.200934   0.566            0.571860    
regionNorth America:govt_health_transform                  -7.577479  17.348659  -0.437            0.662809    
regionSouth Asia:govt_health_transform                     -0.006341   0.196097  -0.032            0.974239    
regionSub-Saharan Africa:govt_health_transform             -0.035198   0.143883  -0.245            0.807028    
regionEurope & Central Asia:private_health_transform        0.666131   0.180003   3.701            0.000287 ***
regionLatin America & Caribbean:private_health_transform    0.065048   0.217825   0.299            0.765580    
regionMiddle East & North Africa:private_health_transform   0.008531   0.243012   0.035            0.972037    
regionNorth America:private_health_transform                3.431308   5.407290   0.635            0.526532    
regionSouth Asia:private_health_transform                   0.502385   0.375367   1.338            0.182498    
regionSub-Saharan Africa:private_health_transform           0.435495   0.165936   2.624            0.009441 ** 
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.6268 on 176 degrees of freedom
  (1 observation deleted due to missingness)
Multiple R-squared:  0.8771,    Adjusted R-squared:  0.861 
F-statistic:  54.6 on 23 and 176 DF,  p-value: < 0.00000000000000022

my.cols <- c("#0a0a0a","#8087A3","#656191","#384c9c","#80808c","#edb880","#ed8618")
my.names <- c("Germay","Poland","France","Italy","United.States","United.Kingdom","Greece")
names(my.cols) <- my.names

mmr_time <- read.csv("mmr_time.csv")
p <-  ggplot(mmr_time, aes(x = Time))

mmr_time

library("tidyverse")
df <- mmr_time %>% 
  rename(
    USA = United.States,
    UK = United.Kingdom
    )

df <- df %>%
  select(Time, Germany, Poland, France,Italy,USA,UK,Greece) %>%
  gather(key = "variable", value = "value", -Time)

head(df)

NA

ggplot(df, aes(x = Time, y = value)) + 
  geom_line(aes(color = variable)) + 
  scale_color_manual(labels=c("France","Germany","Greece","Italy","Poland","United.Kingdom","United.States"),values = c("#0a0a0a","#c8c8cc","#9f9dd1","#4d49b8","#f76902","#edb880","#ed8618")) +
  theme_bw() +
  labs(x = "Year", y = "MMR", color = "Country")

my.cols <- c("#0a0a0a","#8087A3","#656191","#384c9c","#80808c")
my.names <- c("USA", "EUU")
names(my.cols) <- my.names

govt_exp <- read.csv("govt_exp_time.csv")
p <-  ggplot(govt_exp, aes(x = Year))

for (i in 1:2){
  p <- p + geom_line(aes_(y = as.name(names(govt_exp[i+1])), colour = 
colnames(govt_exp[i+1])))
}
p + scale_colour_manual("", 
                        breaks = as.character(my.names),
                        values = my.cols) + labs(y= "Government Health Expenditure Per Capita (Current US $)")
p

my.cols <- c("blue","orange")
my.names <- c("USA", "EUU")
names(my.cols) <- my.names

priv_exp <- read.csv("private_exp_time.csv")
p <-  ggplot(priv_exp, aes(x = Year))

for (i in 1:2){
  p <- p + geom_line(aes_(y = as.name(names(priv_exp[i+1])), colour = 
colnames(priv_exp[i+1])))
}
p + scale_colour_manual("", 
                        breaks = as.character(my.names),
                        values = my.cols) + labs(y= "Private Health Expenditure Per Capita (Current US $)")

p